Genetic Diversity and Phylogenetic Relationships Among Strains of Prevotella (Bacteroides) Ruminicola from the Rumen

Home , Prevotella

INTERNATIONALJOURNAL OF SYSTEMATICBACTERIOLOGY, Apr. 1994, p. 246-255 Vol. 44, No. 2 0020-771 3/94/$04.00 +0 Copyright 0 1994, International Union of Microbiological Societies

Genetic Diversity and Phylogenetic Relationships among Strains of Prevotella (Bacteroides) ruminicola from the Rumen

GORAZD AVGUSTIN,? FRANK WRIGHT,$ AND HARRY J. FLINT* Rowett Research Institute, Bucksburn, Aberdeen AB2 9SB, United Kingdom

A high degree of genetic diversity among 29 strains of Prevotella (Bacteroides) ruminicola from the rumen was revealed by comparing restriction fragment length polymorphisms in 16s rRNA genes, sodium dodecyl sulfate-polyacrylamide gel profiles of total-cell proteins, and G+ C contents of chromosomal DNAs. In order to obtain information on phylogenetic relationships, the sequences of a 389-bp region of the 16s rRNA gene, including variable regions 4 and 5, were compared for 10 strains. These 10 strains formed a single group when their sequences were compared with 16s ribosomal DNA sequences from other species, including Bacteroides spp. from the human colon. On the other hand, the great genetic distances between many P. ruminicola strains, including P. ruminicola subsp. brevis B,4 and GA33 and P. ruminicoh 23T (T = type strain), support the hypothesis that these organisms should be reclassified into new species. We identified signature oligonucleotides based on 16s ribosomal DNA sequences that distinguished strains related to strains 23T, B14, GA33, and M384, as well as an oligonucleotide that specifically recognized all but one of the Bacteroides and Prevotella strains tested. On the basis of the priming activities of these signature oligonucleotides in PCR reactions and on other criteria, we concluded that 12 of the original 29 strains were related to strain 23T, 4 were related to strain B14, and 4 were related to strain GA33. While there are clear grounds for subdividing the species P. ruminicola on the basis of genotypic differences, it is appropriate to delay formal reclassification until further work on the phenotypic differentiation of the new groups is completed.

Prevotellu (Bucteroides) ruminicolu has long been recognized distinct species might be represented by these bacteria. In this as one of the most numerous species inhabiting the rumen (5, study we investigated genetic variation in and phylogenetic 20) and can also account for a high proportion of the hind gut relatedness among P. ruminicolu strains of rumen origin. Our microflora of nonruminants, including pigs and humans (18, results fully support previous evidence indicating that there is 30,31). In a recent study, Van Gylswyk (43) found that as many a high degree of genetic divergence among isolates regarded as as 60% of bacterial isolates from rumina of silage-fed cows P. ruminicolu strains and, together with other recent evidence, belonged to this species. The new genus Prevotellu was recently provide the basis for redefining new species within the group. created (37) to distinguish certain former Bucteroides species, including Bucteroides ruminicolu, Bucteroides meluninogenicus, MATERIALS AND METHODS and Bucteroides orulis, from the “true” Bucteroides species more closely related to Bucteroides fragilis. P. ruminicolu has Strains. The bacterial strains which we used and their generally been regarded as a “widely adapted” species. One origins are shown in Table 1. The isolate of strain 23 which we potential role of this organism is in the degradation of plant used gave total-protein sodium dodecyl sulfate (SDS)-poly- cell wall polysaccharides, such as hemicellulose (7, 47) and acrylamide gel electrophoresis (PAGE) profiles identical to pectin (8). P. ruminicolu strains lack a true cellulase system, are the profiles of ATCC 19189 and is the P. ruminicolu type strain. not able to degrade crystalline cellulose, and do not cause M2 medium (17) was prepared anaerobically under 0,-free extensive solubilization of plant cell wall material in pure CO, by using the methods of Bryant (4). cultures. On the other hand, there is evidence that these DNA extraction. The method used to extract chromosomal organisms contribute to plant cell wall degradation by acting DNA was based on a method described by Ausubel et al. (1). synergistically with cellulolytic bacteria (28). In addition, P. Cultures were grown to the stationary phase in M2 broth in ruminicolu strains can generally utilize cellodextrins (32), 100-ml crimp-sealed bottles under anaerobic conditions at starch (6), and a range of soluble sugars. One of their most 38°C. The cells were centrifuged at 4,000 X g for 10 min and significant roles, however, may be in protein and peptide resuspended in 9.5 ml of TE buffer. A 0.5-ml portion of 10% breakdown (26, 44, 45). SDS and 50 pl of a 20-mg/ml proteinase K solution were It has been shown that P. ruminicolu strains exhibit a high added, and the mixture was incubated for 1 h at 37°C. After degree of genetic divergence. Mannarelli et al. (23) found that addition of 1.8 ml of 5 M NaCl and 1.5 ml of a solution the DNA G+C contents of 14 strains varied between 38 and 51 containing 10% hexadecyltrimethyl ammonium bromide in mol%, while 8 strains exhibited less than 30% relatedness with 0.7% NaCl, the mixture was incubated for 20 min at 65°C and the type strain of the species in total DNA-DNA hybridization then extracted with an equal volume of chloroform-isoamyl experiments. These authors concluded that as many as nine alcohol (24:l). Nucleic acids were recovered from the aqueous phase by spooling after precipitation with 0.6 volume of isopropanol and were purified further by CsC1-ethidium bro- * Corresponding author. Mailing address: Rowett Research Insti- mide density gradient centrifugation. The ethidium bromide tute, Greenburn Road, Bucksburn, Aberdeen AB2 9SB, United King- dom. Phone: (0224) 716651. Fax: (0224) 715349. was removed by sequential extraction with water-saturated t Present address: Zootechnical Department, Biotechnical Faculty, n-butanol, samples were dialyzed extensively against TE buffer University of Ljubljana, Groblje 3, 61230 Domiale, Slovenia. at 4”C, and the DNA was concentrated by ethanol precipita- $ Present address: Scottish Agricultural Statistics Service, Edin- tion. burgh EH9 352, United Kingdom. G+C determinations. DNA base compositions (G+C con-

246 VOL.44, 1994 VARIATION AMONG P. RUMINICOLA STRAINS 247

TABLE 1. Origins and characteristics of the 29 P. rurninicola strains examined in this study

DNA G+C content PCR amplification with the following signature oligonucleotide primersd:I Overall Origin,, SDS-PAGE 16s ribosomal DNA (mol%)' Strain group restriction pattern This group Previous BacPre 23 GA33 B,4 M384 study studies 23T 1 1 49.2 50.6,' 49.4 + A 118B 1 I ND4 50.6,' 50.9 + A TC18 2 2 46.5 + A TC35 2 2 47.7 + A TC44 2 3 48.9 + A TC27 2 3 47.8 + A TS4-6 2 ND 49.9 + A TF1-10 2 4 ND A TF1-5 2 4 ND A TF1-2 2 4 47.0 + A TC2-3 2 47.7 + A TS 1-2 2 ND + A TC2-24 2 48.9 + A TF2-5 2 47.0 + A GA33T (= ATCC 19188T) 5 ND 50.9.' 50.9 + A FC2 6 5 48.5 A FC4 6 5 48.9 + A FC6 6 ND ND A 223lM217 3 ND 49.1 + B TC2-28 2 50.2 + TS2-7 2 ND 46.3 + 9958178 4 45.3 + TC20 2 44.1 - 2202 7 ND 43Sh + M384 6 42.8 43.8" + B B14 8 3 6 40.7 41.6" + B TC1-1 2 3 6 42.1 + B TF1-3 2 3 6 39.2 + B TS1-5 2 3 39.7 + B '' Strains were grouped as far as possible on the basis of SDS-PAGE and 16s ribosomal DNA data (see text). 1, M. Cotta, US.Department of Agriculture, Peoria, Ill. (5); 2, N. 0. Van Cy swyk, Uppsala, Sweden (43); 3, Rowett Research Institute, Aberdeen, United Kingdom (12); 4, H. N. Shah, London Hospital Medical College, London, United Kingdom; 5, American Type Culture Collection, Rockville, Md.; 6, R. J. Wallace, Rowett Research Institute, Aberdeen, Scotland (26,44); 7, National Collection of Food Bacteria, Shinfield, United Kingdom; 8, J. B. Russell, Cornell University, Ithaca, N.Y. (5). The probable error for DNA G+C content determinations is ? 1%. Strains TC20, 2202, M384, B,4, TC1-1, TF1-3, and TS1-5 have G+C contents that are less than 45 mol%. BacPre, 23, GA33, B,4, and M384 are oligonucleotides that were used as forward primers in PCR; these primers were used together with a universal reverse primer (see Fig. 7). ' Data from reference 23. fData from reference 29. R ND, not determined. Data from reference 22a. tents) were estimated by the thermal denaturation method supplied by Boehringer Mannheim, Mannheim, Germany. (24), using 0.15 M NaC1-0.015 M trisodium citrate and a Unincorporated label was removed by spun column chroma- Gilford model 240 spectrophotometer (38). The DNA base tography, and hybridizations were performed at 65°C as de- ratio was calculated with the following equation: G+C content scribed previously (1 1). = 50.9 + 2.44 (melting temperature of unknown DNA - PCR amplification of 16s ribosomal DNA regions V4 and melting temperature of Escherichia coli DNA). E. coli B DNA V5. Chromosomal DNAs from 10 P. ruminicola strains were was used as the standard. used as target DNAs in PCR reactions by using procedures 16s ribosomal DNA restriction fragment length polymor- described by Sambrook et al. (34). Taq polymerase, reaction phism. Chromosomal DNA was digested to completion with buffers, and deoxyribonucleoside triphosphates were obtained EcoRI (Boehringer Mannheim), and fragments were sepa- from Boehringer Mannheim and were used according to the rated in a 0.6% (wt/vol) agarose gel. The DNA fragments were manufacturer's recommendations. Reaction mixtures (100 pl) transferred to a Genescreen Plus hybridization membrane containing approximately 100 ng of target DNA and 200 pmol (Amersham, Little Chalfont, United Kingdom) by Southern each of the forward and reverse primers were heated to 94°C blotting. The DNA used for probe preparation was an equimo- for 5 rnin before Taq polymerase was added. They were then lar mixture of BstEII-Hind111 fragments (approximately 650 subjected to 30 cycles of amplification as follows: denaturation and 950 bp) from plasrnid pKK3535 (3), which represented the at 95°C for 1 rnin (5 rnin in the first cycle), annealing at 47 or majority of the E. coli 16s rRNA gene. The fragments were 51°C for 2 min, and synthesis at 72°C for 3 rnin (10 rnin in the separated on a 0.8% agarose gel, purified by using GeneClean last cycle). DNA products were recovered by chloroform (Bio 101, La Jolla, Calif.), and labelled with ["PIdCTP by the extraction and electrophoresed on horizontal agarose gels. oligonucleotide primer extension method (9, lo), using a kit Cloning and sequencing of PCR products. The PCR prod- 248 AVGUSTIN ET AL. INT.J. SYST.BACTERIOL. ucts were subjected to double digestion with restriction endo- nucleases BamHI and SalI (Northumbria Biologicals, Ltd.) and then cloned into pUC13 (27). E. coli DH5a transformants carrying inserted plasmids were identified on the basis of formation of white colonies on Luria-Bertani agar plates containing 5-bromo-4-chloro-3-indolyl-~-~-galactopyranoside (X-Gal) and ampicillin. DNA sequencing was performed with at least two independent clones by the dideoxy chain termina- tion method (35) from double-stranded templates, using Se- quenase 2 in the presence of 7-deaza-dGTP and M13 forward ( - 40) and reverse sequencing primers (United States Bio- chemical). Internal oligonucleotide primers were also prepared with a Cruachem synthesizer. Sequence analysis and phylogenetic tree construction. Se- quences were analyzed and manipulated by using the GCG (University of Wisconsin) and CLUSTAL V packages avail- able through the Daresbury SEQNET Facility, Daresbury, United Kingdom. The 16s rRNA sequences which were not determined in this study were obtained from the GenBank data base. Sequences were multiply aligned by using the CLUSTAL V package (16) and the default parameters provided for this purpose; the resulting alignment was then checked manually. The painvise distances between pairs of aligned sequences were calculated, and regions which aligned with gaps in the other sequences were ignored. The distances were corrected for multiple substitutions by using Kimura’s two-parameter method (21). This correction had no effect on the final tree topology because the substitution rate in 16s rRNA is low. A neighbor-joining tree (33) was then constructed from the matrix of paired distances. In order to statistically evaluate the tree produced by the neighbor-joining method, the technique known as bootstrapping was used. The original data were resampled 2,000 times as recommended by Hedges (15). The resulting P value given at each node of the tree has an accuracy of k 1%.Thus, a bootstrap P value of 95% at a tree node indicates that 94 to 96% of samples confirmed the node. Bootstrap P values derived from 16s rRNA phylogenetic trees must be treated with some caution (25) because the substantial secondary structure of 16s rRNA violates the bootstrap assumption of independence. FIG. 1. SDS-polyacrylamide gel separations of total-cell proteins SDS-PAGE. SDS-PAGE of P. ruminicola cell extracts was from 27 P. rurninicola strains. Three groups of strains can be distin- performed by the method of Laemmli (22). Cells were grown guished in panel A (see Table 1). The strains included in panel B and to the stationary phase in M2medium, harvested by centrifu- two additional strains (223/M2/7 and TS2-7), which were compared on gation at 2,500 X g in sealed tubes, and washed twice in other gels (data not shown), did not group with any of the other strains. distilled water. After SDS sample buffer (22) was added, samples were boiled for 10 min and then subjected to mild sonication in a Nusonic bath (three times for 60 s, with cooling remaining 17 strains, however, no two profiles were sufficiently on ice). similar to allow certain grouping. Restriction fragment polymorphism in 16s rRNA genes. As an alternative approach to establish strain relationships, we RESULTS examined the patterns obtained when we hybridized cloned fragments of the E. coli 16s rRNA gene to restriction enzyme Total-protein SDS-PAGE comparisons. A total of 29 P. digests of total P. ruminicola chromosomal DNA. This ap- ruminicola strains were examined; 18 of these (those whose proach has recently proved to be valuable for many groups of designations begin with TS, TF, or TC) are strains that were bacteria, including Bacteroides spp. (40). Strains in the three isolated recently from the rumina of silage-fed cows in Sweden groups previously established on the basis of SDS-PAGE by Van Gylswyk (43). Six isolates were from the Rowett comparisons generally, but not invariably, produced similar Research Institute, while the remaining five strains were from EcoRI restriction fragment patterns (Fig. 2 and Table 1). stock collections (Table 1). Three other groups of strains, one of which included type A comparison of cell extracts by SDS-PAGE led to identi- strain 23, exhibited internal similarities in their EcoRI diges- fication of three groups of related strains (Fig. 1). Three recent tion patterns. The remaining 13 strains examined exhibited a Rowett Research Institute isolates produced total-protein pat- remarkable degree of variation in their restriction fragment terns almost identical to the P. ruminicola subsp. brevis type patterns. strain GA33 pattern, while three of the Swedish isolates G+C contents. Mannarelli et al. (23) found that the G+C produced patterns almost identical to the strain B,4 pattern. A contents of the 14 P. ruminicola strains which they examined third group consisted of four similar Swedish strains. For the ranged from 38 to 51 mol%. A similar range was found in this VOL.44, 1994 VARIATION AMONG P. RUMINICOLA STRAINS 249

forward 5’ gaattcgtcgacTGCCAGCAGCCGCGGTAATA3’

EcoRI SaiI

reverse 5 ’ aagcttggatcCCCGTCAATCCfXITGAGTT3’

HidIIBamHI FIG. 3. Oligonucleotide primers used for PCR amplification of a 389-bp region of the 16s rRNA gene from rumen Prevoteffa strains. The bases in lowercase letters are bases that were added to create restriction endonuclease sites for use in cloning amplified fragments. The bases in uppercase letters are complementary to bases in conserved regions of the 16s rRNA gene.

be grouped definitively with it or with each other on the basis of the other criteria used. Chromosomal DNA was prepared from each of the 10 strains and used as a template in amplification reactions with Taq polymerase by using primers that were designed and synthesized to recognize conserved regions some 389 bp apart in the 16s rRNA gene (Fig. 3). The forward primer (residues 516 to 535 in E. coli [see reference 31) was from a region believed to be strongly conserved among eubacteria (46). The reverse primer (residues 908 to 928 in E. coli [3]) was also from a highly conserved region but contained a residue known to be different in E. coli and B. fiagilis, and the B. fragilis sequence was assumed to be more likely conserved in Prevotella strains. The amplified sequence included variable regions 4 and 5. FIG. 2. Hybridization of 32P-labelled DNA fragments carrying E. Each primer included additional residues at the 5’ end, which coli 16s ribosomal DNA sequences to EcoRI-digested chromosomal provided sites for restriction enzyme cleavage to allow cloning DNAs from different P. ruminicola strains. Chromosomal DNAs from of the amplified product into pUC vectors. Not all of the 24 strains and from E. cofi B were completely restricted with EcoRI, strains gave amplified products when an annealing tempera- and the fragments were transferred to a nylon filter after separation on ture of 51°C was used, and therefore a temperature of 47°C a 0.8% agarose gel. Fragments carrying approximately 1,600 bp of the was used routinely; this was almost certainly due to variations E. cofi 16s rRNA genes were used as probes (see Materials and in the sequence recognized by the reverse primer. The se- Methods). The values on the left are the sizes (in base pairs) of markers (1-kb ladder [Bethesda Research Laboratories]). quence of the amplified region was determined on the basis of at least two independent clones for each strain (Fig. 4). Estimates of genetic distances for all 10 P. ruminicola strains and representatives of several other bacterial genera, based on study (Table 1). In agreement with Mannarelli et al. (23), B,4 the data in Fig. 4, are shown in Table 2. Multiple sequence had a G+C content close to 40 mol%, and the G+C content comparisons were also made for the same group of strains, and of type strain 23 was close to 50 mol%. The three strains which a phylogenetic tree was constructed on the basis of the genetic produced SDS-PAGE patterns similar to the B,4 pattern also distance measurements (Fig. 5). Our conclusions are summa- had similarly low G+C contents. Only three other strains had rized as follows. First, the 10 P. ruminicola strains, which were G+C values of less than 45 mol%; the values for the remaining classified as Bacteroides strains until recently, all fall in a broad 17 strains for which data were obtained were between 45 and group that also includes the Bacteroides species Bacteroides 51 mol%. vulgatus, B. frasilis, Bacteroides thetaotaomicron, and Bacte- Comparison of 16s rRNA gene sequences. In order to obtain roides distasonis. All of these organisms possess sequence more precise information concerning the phylogenetic rela- elements that are characteristic of the Bacteroides-Flavobacte- tionships of the strains, we compared the sequences of variable rium phylum defined by Woese (48) (Table 3). The 10 strains regions of their 16s rRNA genes. Ten strains were chosen for form a separate cluster distinct from Bacteroides species (Fig. this study. Strain 23 was included as the current P. ruminicola 5), which supports the hypothesis that they should be reclassi- type strain, on which any narrow definition of the species P. fied in new genera (36, 37). The variation within this cluster is ruminicola should be based (see reference 23). Strains TF1-2, considerable, however. Strains 23T, TC27, TC18, and TF1-2 B,4, and GA33T (T = type strain) were chosen as representa- have almost identical sequences and clearly belong to the tives of the three SDS-PAGE groups described above; we also narrowly defined species P. ruminicola as exemplified by type included strain M384, which appears to typify another SDS- strain 23. On the other hand, the genetic distances between PAGE group, as determined in a separate study (26). B,4 and other strains of P. ruminicola are often greater than the genetic GA33T were previously identified as members of Bacteroides distances between representatives of two different genera of ruminicola subsp. brevis (19). 223/M2/7 was chosen because of gram-positive bacteria (Clostridium perpingens and Bacillus its importance in previous genetic studies (13,39). The remain- subtilis). This is true, for example, for strains GA33” and B,4, ing four strains examined, TC18, TC27, TC2-5 and TC2-24, all which were once regarded as members of the same subspecies. had G+C contents (46.5 to 49 mol%) reasonably close to the The 16s rRNA sequence comparisons revealed an unambig- G+C content of type strain 23 (49 to 50 mol%) but could not uous division between the group of strains represented by B,4, 250 AVGUSTIN ET AL. INT.J. SYST.BACTERIOL. VOL. 44, 1994 VARIATION AMONG P. RUMINICOLA STRAINS 251

FIG. 4. Multiple sequence alignment of 16s ribosomal DNAs from 10 P. rurninicofu rumen strains. Sequences were determined as described in Materials and Methods for both strands of the region shown. A black background indicates that a residue was conserved in at least 6 of the 10 aligned sequences. The region shown corresponds to positions 516 to 907 of the E. cofi 16s rRNA gene (3; also see text).

M384, and 223/M2/7A, which is referred to below as group B, and 1 gave positive results with the M384 primer (Table 1 and and the remaining seven strains, which is referred to below as Fig. 7). The remaining eight strains (2202, 223/M2/7, 9958/78, group A. Group A was unambiguously subdivided into strain TC2-24, TC2-28, TC20, TS2-7, and TF2-5) did not give posi- 23T-like strains (five strains, including the somewhat distantly tive amplification results with any of the four group-specific related organism strain TC2-24) and the strain GA33T-like primers. In addition, we identified an oligonucleotide that strains (GA33T and TF2-5) (Fig. 5). was conserved in all of the Prevotella and Bacteroides strains Signature oligonucleotides. Potential signature sequences in shown in Fig. 5 (Fig. 6). This oligonucleotide, designated the aligned 16s rRNA sequences were identified for strains BacPre, did not occur in the 16s ribosomal DNA of any other 23T, GA33T, B,4, and M384 (Fig. 6). Oligonucleotides having bacterial group examined in our data base searches. When these sequences were used as forward primers in PCR reac- used as a forward primer in conjunction with the universal tions along with a “universal” reverse primer (5’-ACGGGCG reverse primer mentioned above, BacPre gave positive ampli- GTGTGTACAAGGCC) (41). Amplification occurred with fication results with all 25 strains tested except TC20 (Table 1). template DNAs from strains belonging to the appropriate As an additional test of the efficacy of the group-specific group and not with DNAs from strains belonging to other primers, we examined five strains isolated recently from ru- groups. Of the 25 strains tested, 10 gave positive amplification mina (26). Three strains (79/1,79/2, and 52/3), which have been results with the strain 23T primer, 4 gave positive results with reported to produce total-protein SDS-PAGE profiles similar the B,4 primer, 2 gave positive results with the GA33T primer, to the M384 profile, were recognized by the M384 group TABLE 2. Genetic distances for 10 P. ruminicola strains from rumina and other bacterial species based on comparisons of partial 16s rRNA sequences

No. of nucleotide substitutions per positionL'

P. ruminicola 23= P. ruminicola TC18 0.0 P. ruminicola TC27 0.0 0.0 0.0 P. ruminicola TF1-2 0.0 0.0 0.0 P. ruminicola TF2-5 0.052 0.052 0.052 0.052 P. ruminicola TC2-24 0.052 0.052 0.052 0.052 0.043 P. ruminicola GA33T 0.067 0.067 0.067 0.067 0.034 0.058 P. ruminicola B,4 0.130 0.130 0.130 0.130 0.130 0.127 0.120 P. ruminicola M384 0.127 0.127 0.127 0.127 0.124 0.131 0.124 0.111 P. ruminicola 223/M2/7 0.144 0.144 0.144 0.144 0.144 0.144 0.147 0.127 0.074 B. vulgatus 0.180 0.180 0.180 0.180 0.200 0.206 0.174 0.187 0.201 0.197 B. thetaiotaomicron 0.184 0.188 0.055 0.184 0.184 0.184 0.174 0.174 0.176 0.175 0.190 B. fragilis 0.187 0.191 0.064 1 0.187 0.187 0.187 0.179 0.176 0.162 0.175 0.189 0.01 B. distasonis 0.230 0.230 0.230 0.230 0.21 1 0.237 0.218 0.244 0.249 0.134 0.144 0.148 0.265 0.222 0.226 0.219 Cytophaga lytica 0.289 0.289 0.289 0.289 0.277 0.280 0.297 0.294 0.296 0.254 0.284 0.241 0.252 0.309 0.305 Clostridiumperfringens 0.323 0.323 0.323 0.323 0.320 0.333 0.290 0.285 0.305 0.265 Fusobacterium nucleatum 0.360 0.305 0.280 0.288 0.309 0.351 0.233 0.360 0.360 0.360 0.338 0.334 0.347 0.343 0.351 0.300 Bacillus subtilis 0.379 0.3880.360 0.292 0.296 0.322 0.356 0.127 0.241 0.379 0.379 0.379 0.352 0.370 0.334 0.317 0.356 0.317 E. coli 0.383 0.393 0.313 0.305 0.313 0.388 0.207 0.297 0.214 0.383 0.383 0.383 0.370 0.393 0.383 0.322 0.388 0.340 Haemophilus influenzae 0.408 0.356 0.374 0.353 0.407 0.249 0.301 0.249 0.147 0.422 0.422 0.422 0.422 0.412 0.442 0.422 0.393 0.407 0.379 Corrected for multiple changes by the Kimura two-parameter method (see Materials and Methods). VOL.44, 1994 VARIATION AMONG P. RUMINICOLA STRAINS 253

Genetic distance Prevotella ruminiwla - -I Strain 23 5’ ATCTTGAGTGAGITCGATG?G 3’ - d$23, TC27, TC18, TF1-2 I Strain GA33 5’ TCCRGAGTGTATI’CGACGTCAG 3’

Strain M384 5’ TAC’ITGAGTACGCACAACGTAGG 3 --I -- Strain B,4 5’ ACTTGAGTGCACAGGAAGCG 3’ I I 1I B BucteroideslPrevoteLla (BacPre) 5’ GAGTACGCCGGCAACGGTGA 3’ ! I FIG. 6. Oligonucleotides that differentiate P. ruminicolu strains. --I These oligonucleotides were selected from multiple alignments as oligonucleotides that distinguish four groups of P. ruminicolu strains PI73I Bacteroides vulgatus I (23=, GA33T, M384, B,4) or all Bacteroides and Prevotellu strains (BacPre). No corresponding sequences were found in FASTA searches 1 %o Bacteroides fragilis of the whole DNA data base. Used in conjunction with a universal reverse primer (see text), these oligonucleotides produced group- I b5 Bacterodes thetaiotaomicron specific amplification of 16s ribosomal DNA sequences from chromo- Bacteroides distasonis somal DNAs (see Table 1 and Fig. 7).

1 Cytophagalytica Fusobacterium nucleatom B. ruminicola subsp. brevis and B. ruminicola subsp. ruminicola Clostridiumperfringens on the basis of morphology and hemin requirement was largely ‘4, meaningless and proposed that the species ruminicola P. Bacih subtilis 2.5 90 should be limited to strains that are closely related to type Escherichia wli 7.0 strain 23, have G+C contents between 49 and 51 mol%, and 1W 9.3 Haemophilus influenzae are more than 60% related to strain 23 as determined by total-DNA hybridization experiments. They concluded that of FIG. 5. Phylogenetic relationships of the P. ruminicolu strains. The the strains included in this study, M384, B,4, and GA33T values in italics are the genetic distances (expressed as number of differed enough in G+C content or DNA-DNA hybridization substitutions per 100 nucleotides) between strains based on a compar- characteristics from each other and from strain 23T that they ison of a 389-bp region of 16s ribosomal DNA including regions V4 should be placed in different species. The painvise DNA-DNA and V5 (Fig. 4; see text). The length of the horizontal lines are proportional to genetic distances; the lengths of the vertical lines are hybridization values for strains 23T, M384, B,4, and GA33T all arbitrary. The values in boldface type are the percentages of trials in ranged from 19 to 22%, whereas the value for 23T and 118B which the branch points indicated occurred when repeated bootstrap- was 83% (23). Since the potential species were represented by ping procedures were performed. single strains, however, it was difficult to reclassify them. On the basis of our comparison of 16s ribosomal DNA sequences and on the basis of recognition by signature oligo- primer, while the other two strains (92/1 and 92/2), which have nucleotides, together with other criteria, we assigned 18 of the been reported to resemble B,4, were recognized by the B,4 29 strains which we studied to one broad genotypic group, group primers. group A, and 6 strains to a second broad genotypic group, group B, which left the status of five strains (2202, TC20, 9958/78, TC2-28, and TS2-7) uncertain (Table 1). As noted DISCUSSION previously (42), strain 2202, which is distributed by the Na- Mannarelli et al. (23) concluded previously that P. rumini- tional Collection of Food Bacteria (Shinfield, United King- cola is genetically a highly diverse taxon. These authors showed dom) as strain 23, is clearly distinct from P. ruminicola 23T; this that the previous subdivision of B. ruminicola into subspecies strain has been used in some genetic and physiological studies

TABLE 3. Comparison of sequences of P. ruminicolu strains at signature positions which characterize the Bacteroides-Flavobucteriumphylum“

Position Bacteroides Ancestral 23T B14 GA33T M384 TC2-24 TF2-5 223IM217A of base” SPP. position‘

~ 564 U U U U U U U 570 U U U U U U U 684 U U U U U U U 698 G G G G G G G 718 U U U U C C C 809 A A A A A A A 812 G G G G G G G 837 C C C U C C G 849 ------868 C C C C C C C 870 U U U U U U U 906 G G G G G G G See reference 14. The sequences of strains TC27, TC18, and TF1-2 (data not shown) were identical to the sequence of strain 23T at these positions. ” Position number in the E. coli sequence (3). ‘’Ancestral cornposition was defined as the dominant composition in the majority of bacterial phyla (14). -, deletion compared with E. coli sequence in the region between positions 832 and 850 in the majority of aligned sequences. 254 AVGUSTIN ET AL. INT. J. SYST. BACTERIOL.

It will also be important to establish whether nonrumen isolates (e.g., isolates from pig guts) regarded as P. ruminicolu strains correspond to any of the new species recognized mentioned above or whether these strains represent entirely different groups. In the future, more rapid and extensive surveys of strains should be possible with the signature oligonucleotides designed in this study.

ACKNOWLEDGMENTS

We are most grateful to N. 0. Van Gylswyk €or generously providing most of the strains used in this study; we are also grateful to Haroun Shah, Saheer Gharbia, and Paul Lawson for valuable discussions, for providing cloned 16s rRNA genes for restriction fragment analyses, and for assistance with G+C content determinations. We also thank R. J. Wallace, M. Cotta, J. B. Russell, and H. N. Shah for providing strains. This work was supported by the Scottish Office Agriculture and Fisheries Department. FIG. 7. Specific amplification of an approximately 800-bp region of 16s ribosomal DNA by using the B,4 oligonucleotide together with the REFERENCES universal reverse primer. Lanes of an agarose gel were loaded with PCR mixtures from experiments performed with chromosomal DNAs 1. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. G. from P. rurninicola B,4 (lane 2), TC1-1 (lane 3), TS1-5 (lane 4), TF1-3 Seidman, J. A. Smith, and K. Struhl (ed.). 1987. Current protocols (lane 5), 92/1 (lane 6),92/2 (lane 7), TC18 (lane S), M384 (lane 9), and in molecular biology, vol. 1. John Wiley and Sons, New York. GA33T (lane 10) and E. coli B (lane 11). Strains 92/1 and 92/2 are 2. AvguStin, G., H. J. Flint, and T. R. Whitehead. 1992. Distribution described in reference 26; other strains are listed in Table 1. Lane 1 of xylanase genes and enzymes among strains of Prevotella (Bac- contained molecular size markers (l-kb ladder [BRL]). teroides) rurninicola from the rumen. FEMS Microbiol. Lett. 99: 137-144. 2a.AvguStin, G., et al. Unpublished data. 3. Brosius, J., T. J. Dull, D. Sleeter, and H. F. Noller. 1981. Gene (12, 47). We propose that the species P. ruminicolu should be organisation and primary structure of a ribosomal RNA operon redefined more narrowly so that it includes only those strains from Escherichia coli. J. Mol. Biol. 148:107-127. likely to be recognized by the strain 23T signature oligonucle- 4. Bryant, M. P. 1972. Commentary on the Hungate technique for otide. On this basis 12 of the strains which we studied would culture of anaerobic bacteria. Am. J. Clin. Nutr. 25: 1324-1328. 5. Bryant, M. P., N. Small, C. Bouma, and H. Chu. 1958. Bacteroides belong to this species; the G+C contents of these strains range rurninicola n. sp. and Succinirnonas amylolytica, the new genus and from 46 to 51 mol%, which is a slightly wider range than the species. J. Bacteriol. 76:15-23. range proposed by Mannarelli et al. (23). The four strains 6. Cotta, M. A. 1988. Amylolytic activity of selected species of related to GA33T were clearly distinct from the strains in ruminal bacteria. Appl. Environ. Microbiol. 54:772-776. group A related to strain 23T, and we believe that these four 7. Dehority, B. A. 1966. Characterization of several bovine rumen strains should form a separate species created by elevating the bacteria isolated with a xylan medium. J. Bacteriol. 91:1724-1729. P. rurninicola subsp. brevis strains related to GA33T to the 8. Dehority, B. A. 1969. Pectin-fermenting bacteria isolated from the species level. The two remaining group A strains, TC2-24 and bovine rumen. J. Bacteriol. 99189-196. TF2-5, are probably best regarded as strain 23T-like and strain 9. Feinberg, A. P., and B. Vogelstein. 1983. A technique for radiola- belling DNA restriction endonuclease fragments to high specific GA33T-like, respectively. activity. Anal. Biochem. 132:6-13. The finding that three recently isolated strains from Sweden 10. Feinberg, A. P., and B. Vogelstein. 1984. Addendum. Anal. Bio- (Table 1) and two strains from Scotland (26) are genotypically chem. 137:266-267. closely related to strain B,4 (presently regarded as a P. 11. Flint, H. J., C. A. McPherson, G. AvguStin, and C. S. Stewart. rurninicolu subsp. brevis strain) suggests that it is appropriate to 1990. Use of a cellulase-encoding gene probe to reveal restriction designate an additional species characterized by low G + C fragment length polymorphisms among ruminal strains of Bacte- contents (39 to 42 mol%). The genotypic differences among roides succinogenes. Curr. Microbiol. 20:63-67. B,4, M384, and 223/M2/7 (group B) were too large for all three 12. Flint, H. J., and C. S. Stewart. 1987. Antibiotic resistance patterns strains to be considered members of the same species, and and plasmids of ruminal strains of Bacteroides rurninicola and Bucteroides multiacidus. Appl. Microbiol. Biotechnol. 26:450-455. separate species status may be appropriate for M384 and 13. Flint, H. J., A. M. Thomson, and J. Bisset. 1988. Plasmid- 223/M2/7 once additional related strains have been character- associated transfer of tetracycline resistance in Bacteroides rumini- ized. In the case of M384, a signature oligonucleotide has cola. Appl. Environ. Microbiol. 54555-560. already confirmed its close relationship with three strains 14. Gherna, R., and C. R. Woese. 1992. A partial phylogenetic analysis recently isolated from rumina (26). of the “Flavobacterium-Bacteroides” phylum: basis for taxonomic Although we believe that there is now compelling evidence restructuring. Syst. Appl. Microbiol. 15513-521. for the creation of at least three new genotypic species for P. 15. Hedges, S. B. 1992. The number of replications needed for rurninicolu strains, based on strains 23T, GA33T, and B,4, we accurate estimation of the bootstrap P value in phylogenetic believe that it is premature to propose naming these taxa studies. Mol. Biol. Evol. 9:366-369. 16. Higgins, D. G., R. Fuchs, and A. Bleasby. 1992. CLUSTAL V: a formally as new species until convenient phenotypic criteria new multiple sequence alignment program. CABIOS 8:189-191. can be provided to distinguish them. There are already indi- 17. Hobson, P. N. 1969. Rumen bacteria. Methods Microbiol. 3B:133- cations that the genetic differences are reflected in functional 149. differences with respect to fermentation patterns, peptide or 18. Holdeman, L. V., I. J. Good, and W. E. C. Moore. 1976. Human protein breakdown, and polysaccharidase activities (2, 2a, 26). fecal flora: variation in bacterial composition within individuals VOL.44, 1994 VARIATION AMONG P. RUMINICOLA STRAINS 255

and a possible effect of emotional stress. Appl. Environ. Microbiol. cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Labo- 3k359-375. ratory Press, Cold Spring Harbor, N.Y. 19. Holdeman, L. V., R. W. Kelly, and W. E. C. Moore. 1984. 35. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequencing Bacteroides, p. 604-631. In N. R. Krieg and J. G. Holt (ed.), with chain terminating inhibitors. Proc. Natl. Acad. Sci. USA Bergey’s manual of systematic bacteriology, vol. 1. Williams and 745463-5467. Wilkins, Baltimore. 36. Shah, H. N., and M. D. Collins. 1983. Genus Bacteroides: a 20. Hungate, R. E. 1966. The rumen and its microbes. Academic Press, chemotaxonomical perspective. J. Appl. Bacteriol. 55403-416. Inc., New York. 37. Shah, H. N., and M. D. Collins. 1990. Prevoteffa, a new genus to 21. Kimura, M. 1983. The neutral theory of molecular evolution. include Bacteroides melaninogenicus and related species formerly Cambridge University Press, Cambridge. classified in the genus Bacteroides. Int. J. Syst. Bacteriol. 40205- 22. Laemmli, U. K. 1970. Cleavage of structural proteins during the 208. assembly of the head of bacteriophage T4. Nature (London) 38. Shah, H. N., R. A. D. Williams, G. H. Bowden, and J. M. Hardie. 227:680-685. 1976. Comparison of the biochemical properties of Bacteroides 22a.Mannarelli, B. M. Personal communication. melaninogenicus from human dental plaque and other sites. J. 23. Mannarelli, B. M., L. D. Ericsson, D. Lee, and R. J. Stack. 1991. Appl. Bacteriol. 41:473492. Taxonomic relationships among strains of the anaerobic bacte- 39. Shoemaker, N. B., G.-R. Wang, and A. A. Salyers. 1992. Evidence rium Bacteroides ruminicola determined by DNA and extracellular for natural transfer of a tetracycline resistance gene between polysaccharide analysis. Appl. Environ. Microbiol. 57:2975-2980. bacteria from the human colon and bacteria from the bovine 24. Marmur, J., and P. Doty. 1962. Determination of the base rumen. Appl. Environ. Microbiol. 58: 1313-1 320. composition of deoxyribonucleic acid from its thermal denatur- 40. Smith, C. J., and D. R. Callihan. 1992. Analysis of rRNA ation temperature. J. Mol. Biol. 5:109-118. restriction fragment length polymorphisms from Bacteroides spp. 25. Marshall, C. R. 1992. Substitution bias, weighted parsimony and and Bacteroidesfragifis isolates associated with diarrhea in humans amniotic phylogeny as inferred from 18s rRNA sequences. Mol. and animals. J. Clin. Microbiol. 30806-812. Biol. Evol. 9:370-373. 41. Stahl, D. A., B. Flesher, H. R. Mansfield, and L. Montgomery. 26. McKain, N., R. J. Wallace, and N. D. Watt. 1992. Selective 1988. Use of phylogenetically based hybridization probes for isolation of bacteria with dipeptidyl amino peptidase type I activity studies of ruminant microbial ecology. Appl. Environ. Microbiol. from the sheep rumen. FEMS Microbiol. Lett. 95169-174. 54:1079-1084. 27. Messing, J. 1983. New M13 vectors for cloning. Methods Enzymol. 42. Thomson, A. M., H. J. Flint, M. Bechet, J. Martin, and H.-C. 101~20-78. Dubourguier. 1989. A new Escherichia co2i:Bacteroides shuttle 28. Osborne, J. M., and B. A. Dehority. 1989. Synergism in degrada- vector, pRRI207, based on the Bacteroides ruminicola plasmid tion and utilization of intact forage cellulose, hemicellulose, and replicon pRRI2. Curr. Microbiol. 24:49-54. pectin by three pure cultures of rumen bacteria. Appl. Environ. 43. Van Gylswyk, N. 0. 1990. Enumeration and presumptive identifi- Microbiol. 55:2247-2250. cation of some functional groups of bacteria in the rumen of dairy 29. Reddy, C. A., and M. P. Bryant. 1977. Deoxyribonucleic acid base cows fed grass silage-based diets. FEMS Microbiol. Ecol. 73:243- composition of certain species of the genes Bacteroides. Can. J. 254. Microbiol. 23:1252-1256. 44. Wallace, R. J., and M. L. Brammall. 1985. The role of different 30. Robinson, I. M., M. J. Allison, and J. A. Bucklin. 1981. Charac- species of bacteria in the hydrolysis of protein in the rumen. J. terization of the cecal bacteria of normal pigs. Appl. Environ. Gen. Microbiol. 131:821-832. Microbiol. 41:950-955. 45. Wallace, R. J., and N. McKain. 1991. A survey of peptidase activity 31. Robinson, I. M., S. C. Whipp, J. A. Bucklin, and M. J. Allison. in rumen bacteria. J. Gen. Microbiol. 137:2259-2264. 1984. Characterization of predominant bacteria from the colons of 46. Weisburg, W. G., S. M. Barns, D. A. Pelletier, and D. J. Lane. normal and dysenteric pigs. Appl. Environ. Microbiol. 48:964-969. 1991. 16s ribosomal DNA amplification for phylogenetic study. J. 32. Russell, J. M. 1985. Fermentation of cellodextrins by cellulolytic Bacteriol. 173:697-703. and non-cellulolytic rumen bacteria. Appl. Environ. Microbiol. 47. Williams, A. G., and S. E. Withers. 1982. The production of plant 49572-5 76. cell wall polysaccharide-degrading enzymes by hemicellulolytic 33. Saitou, N., and M. Nei. 1987. The neighbour-joining method: a rumen bacterial isolates grown on a range of carbohydrate sub- new method for reconstructing phylogenetic trees. Mol. Biol. Evol. strates. J. Appl. Bacteriol. 52:377-387. 4:406-425. 48. Woese, C. R. 1987. Bacterial evolution. Microbiol. Rev. 51:221- 34. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular 271.